Nomogram model to predict postoperative infection after mandibular osteoradionecrosis surgery

Abstract

Abstract Osteoradionecrosis of the mandible (ORNM) is one of the most dreaded complications of radiotherapy. The poor healing capacity of soft tissue after radiation may lead to surgical failure. The current study was designed to identify prognostic factors for postoperative infection (PPI) and propose corresponding prophylaxis and intervention protocols. A retrospective study was conducted concerning ORNM patients from 2000 to 2015. A risk-stratification score and nomogram model were established to predict the risk of PPI. A total of 257 patients were analyzed, and the total incidence of PPI was 23.3% (60/257). In multiple logistic regression analysis, radiation dose $$\geqslant $$ ⩾ 80 Gy (versus <80 Gy, OR = 2.044, P = 0.035, 95% CI: 1.05–3.979), bilateral ORNM (versus unilateral, OR = 4.120, P = 0.006, 95% CI: 1.501–11.307), skin fistula (versus none, OR = 3.078, P = 0.040, 95% CI: 1.05–9.023), and implant utilization (versus none, OR = 2.115, P = 0.020, 95% CI: 1.125–3.976) were significantly associated with PPI. The susceptibility to PPI in patients with risk-stratification scores of 14–22 was 2.833 times that of patients with scores of 7–13, and 7.585 times that of cases defined as scores of 0–6. The discrimination capability of the nomogram model was estimated using a ROC curve with an AUC of 0.708, revealing potentially useful predictive abilities. In conclusion, current risk-stratification scores and nomogram models effectively predicted the risk of PPI in ORNM patients.